Study on Intelligent Classification of Aging Heat-Resistant Materials.

Area of Science:

• Materials Science
• Metallurgy
• Artificial Intelligence

Background:

• High chromium martensitic heat-resistant steel is crucial for next-generation incinerator pressure components.
• Long-term service leads to material aging, impacting safety.
• Accurate aging identification is vital for power plant operational safety.

Purpose of the Study:

• To develop an automatic aging grading model for high chromium martensitic heat-resistant steel.
• To leverage deep residual networks and multiscale metallographic data for accurate aging assessment.
• To enhance the safety and reliability of power plant components.

Main Methods:

• Development of a deep residual network-based automatic aging grading model.
• Construction of a multiscale dataset using image reduction techniques.
• Training and validation using feature pyramid network models.
• Classification of P91 steel microstructure images at different scales.

Main Results:

• The proposed model demonstrates high accuracy in identifying aging states from metallographic data.
• Multiscale datasets significantly improve the model's performance.
• Feature pyramid network training further enhances prediction accuracy.
• The deep learning method effectively classifies damage and aging in P91 steel.

Conclusions:

• The deep learning model provides a valid and accurate method for assessing the aging of high chromium martensitic heat-resistant steel.
• This approach is crucial for ensuring the long-term service safety of power plant components.
• The study validates the effectiveness of AI in material science for industrial applications.